Sep 08, 2019
The capstone project has done a fantastic job of drilling in some of the important fundamentals taught in the rest of the courses. It is definitely worth taking if you've done the rest of the courses.
Jun 14, 2017
Good course, some of the assignments could have been more explicit with expectations in cases where specific implementation details matter but the forums were helpful in that regard anyways.
by IURII B•
Aug 07, 2017
by Francisco J D d S F G•
May 24, 2018
I really enjoyed this course - it goes from the very basics of Spark and tests your ability regarding Scala knowledge.
The capstone project is really well structured, but at times it felt like the instructors could give more information concerning some definitions around the project, i.e. how all of the case classes from the models file interconnect or relate to each other. By the end of the project it would be helpful explaining how the Layer class was actually structured as well.
In reality my opinion of this course is 4.5 out of 5 - I think the instructors made an admirable job trying to convey most of specialization concepts into the capstone, so congratulations :-). If not for some of the aspects above, I would rate it 5/5.
by Edem N•
Jul 29, 2019
A good project broken down into stages that can be independently tested to produce a final product that is both complex and interesting. The forum didn't have any of the TA's responding during my period and as such mean't that any concerns that I had were not being addressed. I had particular concerns with the grader input data where I had issues with its output. Apart from that a great way to put all the skills of the Scala Specialization into use.
by Niels P•
Jul 24, 2017
Excellent capstone that lets you test your knowledge of Scala in a (semi)real project environment. Though i found that the main difficulty for me was in understanding the context of geographical computations. Making it difficult to find if not passing test was due to error in my code or in my understanding of geographical context.
by Rob S•
Nov 26, 2018
Interesting project and fun to work on.
As with many of the other courses, could benefit from providing a few tests. Even after writing property-based tests using ScalaCheck there were some grader results that I was unable to reproduce.
In the end, still worth the time and effort invested.
by Alan A•
Nov 16, 2019
The course is great, but if you pass all tests, in Week 4 you can get the certificate. The project is linked to Week 1, which is weird because doesn't match well with the schedule set in Coursera.
by Rodolfo N P•
Aug 09, 2018
For a lot of sections one couldn't really use Spark, said sections were restricted to translating Spark datastructures to Scala collections and working with those.
by Chet W•
Jan 29, 2018
Very cool to see the entire application fit together eventually. However, feel sometimes the explanation for the exercise could have used more details/examples.
by Jean-Francois T•
Apr 18, 2017
A bit sad that actual full working solution is not graded as it invovled generating files on disk, which is compuation intensive...
Jun 06, 2018
Good Spark / Big Data project.
Provides a good example of real problem project focussed on Spark and big data analysis.
by Алексей Ш•
Aug 03, 2017
The tasks are very interesting! Unfortunately, sometimes the causes for the failure of the tests are not clear.
by Natalija I•
Jan 04, 2019
Capstone project is challenging specially because of the performance standards. I enjoyed solving the problems
Jun 16, 2017
Interesting project, but exercises were too simple. I was able to complete with course in 5 evenings.
by Carsten I•
Oct 20, 2019
Tough but interesting, it definitely showed you the importance of writing unit test!
Apr 18, 2017
Good project enjoyed playing with the Scala.ja
by Wolfgang G•
Sep 01, 2017
pretty tough, but informative!
by Hong C•
Apr 19, 2020
Well designed project.
by Michael L•
Jan 11, 2020
I did the original courses back in 2016 and have used Scala consistently in various roles. I've done so many courses, I wanted to just finish this one and complete the specialization. I think I'm familiar with various Typelevel projects, SBT itself (I wrote a cucumber plugin for it) and have used Shapeless and Cats in anger. I'm familiar with ScalaCheck and have done a few O/S projects. This course ... perhaps I'm being unfair, but the project doesn't feel 'idiomatic Scala'. I think the course could use some of the Typelevle libraries as a starting point and lead devs into how to write functional Scala. In my experience - Scala is used a) for a functional programming language b) to drive Spark. Just as people will use Python for numpy/tensorflow (e.g. you can't compare Spire with numpy). The ML niche is lost, so I hope the content of these courses will still be useful for another 5 years.
by Ryan S•
Jun 12, 2018
I enjoyed this course, but I believe there are some errors in the grading code related to interpolating colors. I raised an issue about it on the week 5 forum but never got a response. The project was interesting, but it would have been nice if there was a way that we could have been a bit more free in how we designed the interfaces. If you were trying to use something like a spark RDD it made it pretty cumbersome, having to always fit everything to an Iterable at the end. ALSO, I tried to add a library when I was doing this to add the shapeless library to this to add some interesting features which would have made the code a lot more general. I spent days getting it to work, only to realize you library dependencies are not actually imported when you submit, it only runs with the base ones that were pre-defined. A warning about that would have been nice.
by Fleur K•
Nov 25, 2017
The first couple of assignments I found harder than the rest, and also the best bits. I found myself enthusiastically telling my colleagues about the data files and the spark code to join them. After that it got a bit too simple perhaps. And a little contrived. Some of the curried functions you had to fill in were way different from what I'd have written if I'd been staring at an empty screen and even after the course I've still not learned why I'd want to use this setup. The Signal bit in the last assignment was just plain silly, I think. It remains completely unclear why you'd want to use Signals here, other than that you want to refer to materials from the reactive programming course, and the sample code for the update doesn't actually work because the update method is protected.
by Роман В•
Jun 24, 2018
Milestones 1-3 was OK but milestone 4 require a much more efficient solution. And the only info grader gives you is the message about timeout. It would be better if requirements for milestones 1-3 were harder to fullfill or milestone 4 would not depend on previous code. But it depends and if your solution is inefficient it's very painful to search the problem all around the code. It's offered to use spark and other tools in this project but using spark is also painful. I get inconsistent result from grader with occasional OutOfMemory exceptions, but this project cries to be implemented with spark. I've learned something during this course but it was too stressfull and painful.
Jun 06, 2020
I'm sad to say that the last course in the specialization was the worst. The grader functions made it quite cumbersome to use Spark, which the test of the specialization was about.
You cannot use Spark throughout the course, because of the grader functions. You have to jump in and out of the Spark domain, which hurts performance and will not allow you to write a complete Spark program.
by Iurii S•
Nov 20, 2017
Would have liked more tests provided with the capstone to supplement often vague specifications. Feedback from autograder is often obscure as well as it refers to some input parameters of the randomized test - hard to take guidance in code improvement from it.
by David P•
Jun 16, 2017
Tough to design a capstone for a course, but I was never really able to get excited about the capstone project. It felt tedious. Not sure if this is the course organizers fault thought...
by Jędrzej B•
May 22, 2020
More about number crunching than fp. The dataset is within the jar, which caused a lot of problems and excluded efficient solutions (and the course is about efficiency as well).